The subject of this patent application relates generally to artificial intelligence, and more particularly to systems and methods for creating and implementing an artificially intelligent agent or system.
By way of background, since the development of the computer, human beings have sought to construct computers that are capable of thinking, learning and carrying on intelligent conversations with humans—in other words, “artificial intelligence.” Some development of such artificially intelligent computers has focused on developing computers that are capable of conversing. Thus, a key area in developing an artificially intelligent computer has been developing a language that allows a computer to process inputs received from humans and to respond with an appropriate and cogent output. One such language is known as Artificial Intelligence Markup Language (“AIML”).
AIML is interpreted and processed by an AIML interpreter, such as Artificial Linguistic Internet Computer Entity (“ALICE”). The AIML interpreter is designed to receive an input from a user and determine the correct response using knowledge encoded in AIML and stored in an AIML knowledge base. In arriving at a response for a particular input, the AIML interpreter searches a list of categories within the AIML knowledge base. Each category contains a pattern that is linked to a single response template. The AIML interpreter matches the user input against the available patterns in the AIML knowledge base. After finding a match in a pattern, the pattern's corresponding response template is activated and a series of actions are carried out by the AIML interpreter.
The known prior art methods for creating such a computer personality generally consist of manually creating and editing that knowledge base and associated response templates (often referred to as “question-response pairs” or “QR pairs”). As such, the process of creating a computer personality having a relatively high level of artificial intelligence can be very labor intensive and can take thousands or even tens of thousands of hours in order to form a believable personality. Furthermore, depending on the particular context in which a given computer personality is to be utilized (i.e., in the medical field, engineering field, general consumer field, etc.), each discrete computer personality may require a unique set of QR pairs. Thus, there is a need for systems and methods for automating the process of creating an artificially intelligent computer personality that is tailored for a desired context.
There are many type of artificial neural networks known in the prior art. Forward passing neural networks faced the problem of not being able to handle XOR logic problems. Later back propagating networks were developed. Recently a problem which relates to all of these inventions has emerged in the form of a blind spot.
Additionally, among the drawbacks found in many prior art systems is a dependence upon grammar and punctuation in order to recognize elements within a sentence. This presents an insurmountable drawback when attempting to adapt these systems to environments where voice recognition rather than text is the input device. Other problems that exist in systems representative of the current art include a lack of flexibility. Because these systems are issued as standards, they are rigid for the time period that a particular version is operational. This makes it very difficult for them to be adapted to changing technological environments as they are encountered. Implementing upgrades involves issuing a new version which gives rise to versioning problems and very often necessitates entire systems being forced to come offline while they are being adapted to a newer version. Other problems include object representation and the need for a simple way to represent any object known or unknown which might be encountered by an artificially intelligent agent or system.
Many attempts to create a standardized object representation format are known in the prior art. One of the more prominent of these is OWL. All of them have a problem which has sparked one of the more rigorous debates in the field of artificial intelligence: Is an artificially intelligent agent or system truly intelligent, or is its intelligence just an extension of the programmer's intelligence? To some degree they attempt to identify objects and store them according to a pre-determined classification set. This gives any artificially intelligent agent or system using the ontology what is necessarily a view of the world as seen through the eyes of the programmer which created the classification, and further any artificially intelligent agent or system using the ontology would have the same view.
Furthermore, in the context of artificially intelligent systems designed for personal use, such as on smart phones and other mobile devices, such prior art systems typically suffer various drawbacks including limited or no protection for personal data which would include a perceived lack of controllability by the user of how personal data is used by the company hosting the artificial intelligence. There have been numerous attempts to secure personal data that is acquired, stored and later accessed by an artificially intelligent agent or system functioning as a personal agent. To date all of these have failed to some degree. Another notable problem is that when multiple users access a single device, mobile or otherwise, they are presented with a single personality. Still another notable problem is the fact that each device owned by a single individual has its own artificial intelligence—in other words, certain elements such as personal information are duplicated and are not transferable between devices. Many attempts at developing an artificially intelligent personal agent are known in the prior art. Some of the most well known include SIRI and Cortana. These suffer from several problems. One such problem is that they do not share a common information base between devices. In addition, certain aspects of an artificial general intelligence (“AGI”) used for human interaction such as voice should be consistent between devices. In other words a given personal assistant should have the same voice from device to device and should have access to and data generated on a particular device when the user access the agent from a different device. This might best be termed a “roaming personality.” Still other problems center on authentication methods for personal data access.
Yet another problem that arises in the context of smart phones and other mobile devices (hereinafter referred to generally as “mobile devices”) is related to advertising. Prior to the growth in popularity of mobile devices, consumers would typically use personal computers as their primary means of accessing the Internet. Given the relatively larger screens and greater computing power of most personal computers (as compared to that of mobile devices), the type, quality and amount of advertising content that can be served to users via personal computers tends to be relatively better than advertising content served via mobile devices. Thus, as a result of consumers using mobile devices increasingly more than personal computers, the effectiveness of Internet-based advertising campaigns appears to be decreasing. Not only that, but consumers are increasingly using their mobile devices as their primary source of news and entertainment, as opposed to traditional news and entertainment sources (i.e., television, newspapers, magazines, etc.), thereby also decreasing the effectiveness of advertising campaigns disseminated through such traditional news and entertainment sources—sources that are also often capable of providing relatively better advertising content as compared to mobile devices. Furthermore, given that consumers are relying more and more on voice commands and text-to-speech applications when using their mobile devices (such that they aren't actually looking at their mobile devices as often), the likelihood of actually viewing advertising content via mobile devices is progressively decreasing. Accordingly, there continues to be a need for effective alternatives for delivering advertising content to consumers.
Aspects of the present invention are directed to solving all of these problems by providing systems and methods for creating and implementing an artificially intelligent computer personality, as discussed in detail below.
Applicant(s) hereby incorporate herein by reference any and all patents and published patent applications cited or referred to in this application.